New Project

jdlech

Member
2007-12-25 7:51 pm
Greetings, and sorry in advance for the huge post.

I did a few small projects back when I was a teen in the '80s but I haven't done anything since. Wow, the DIY speaker biz has certainly grown up since then. I've been obsessing over catching up for about 2 months now. I've looked around several commercial stores (and the internet) that sell speakers and am emphatically unimpressed with what's commercially available. It seems that the price is so jacked up that you only get half of what you should for your money. So, I decided it's time to roll up the sleeves and do it myself.

First, I'm a long time DIYer and am an experienced metal fabricator. I know my way around a shop, so this is not a daunting project to me. The most daunting aspect for me is actually the veneer and finish; which I'm totally without experience. I even considered welding up some giant aluminum cans for speaker boxes, but I would have to weld a million braces in it and it would probably still sound like a giant tin can. :eek:

My reason for posting here is mainly for comment and criticism (the project, not me). I still consider nothing cast in stone and probably won't be until I've ordered parts and started cutting the wood. This represents a significant investment for me so I would like to hear from those who know better before I open up the wallet.
Here's what I have so far.

Drivers:
4ea. - Dayton SD315-88 12" Shielded DVC Subwoofers
Parts Express
Each will be wired in series - 16ohms
All 4 will be wired in series-paralell - 16 ohm average
All 4 will be in an isobaric chamber in push pull configuration.

2ea. Dayton RS52AN-8 2" Dome Midrange
Parts Express
Wired in series

1ea. Tang Band 25-1372SC 1" Titanium Tweeter
Parts Express - again

First comment: It's quite hard to find T/S parameters for midranges and tweeters, let alone actual measurments. Parts Express got a real leg up on any other distributer because they publish it. I've been using Bassbox pro and X-over pro to do all my modelling. I found that I can import the CLIO freq. response directly from their website to my program - sweet!

A note on my woofers/configuration: I was inspired by James Lehman's Twin ISO 8s . But I modelled nearly everything offered by MCM (within a reasonable price for this project) and they all came up lacking in either low end response or overall sensitivity. I think I modelled about 40 woofers for this project so far. The problem with repeating his project is that he used predicted response rather than actual measurements. No speaker comes close to their predicted response curve - especially as they approach their Fs and the breakup region. I'm leery of using predicted response curves. Luckily, Parts Express provides actual measurments for most of their woofers. MCM offers a number of graphs, but I had to sit there for hours manually typing in approximations based upon them - not what I consider fun. And they all seriously deviated from their T/S predictions where it counted most. Once I hit upon these 12" dual coil subs (and their actual measured response curve), the models just sort of fell into place as better than all the rest. And my models put the woofer impedance between 8 and 16 ohms (freq. dependent) which I think is the "golden zone" to produce clean tight bass without overworking the amp, according to the opinion of the author of this article to which I happen to agree based upon my albeit subjective experience.

My choice of midrange drivers actually show what I'm talking about. Their T/S parameters indicate they are actually 6 ohm drivers, not 8. This makes them power hungry and if left to their own devices, would actually reduce the output of both the woofers and the tweeter. Xover Pro predicts an overall 3 db loss across the board if I only use 1 midrange because of a low overall impedance. Even with them wired in series, the lowest impedance is about 6.25 ohms @ 1150Hz. A third driver improves this slightly (overall, about 1/2 db.) certainly not enough to justify the extra cost, IMO.

As it stands, using actual measurments for the woofers and the midrange, and my own approximatons based on the tweeters response graph, my SPL @ 1W. should be somewhere in the neighborhood of 101 db. But if what I suspect is true, the combination of box response and woofer response is additive and the whole will probably be quite a bit less. I'll be happy with 91db @1w/1m. since I'll be driving it with a 220w amplifier. That should be plenty of headroom for dynamic range and the higher volumes I require. I have to admit that I'm no spring chicken anymore and between jet aircraft engines, loud music as a teen and tween and loud shop environs, my hearing has taken a beating over the years. I have tinnitus now and my hearing beyond 14K is about trash. But against all logic and reason, old audiophiles never give up - or in.
 

jdlech

Member
2007-12-25 7:51 pm
Part 2...

Crossover:
X-over pro does not model cascading crossovers, so I would be interested if anyone can equal or exceed the response graph I got using a cascading xover. The xover frequencies are not tied to each other right now anyway. So I'm not certain if I could do a cascading xover at all. but as I have essentially no experience with them, I wouldn't know unless someone tells me. :confused: My crossover points are about 3 1/4 octaves apart. Here's an interesting article that mentions why it might be important - read conclusion #2.

Low pass filter - 250Hz w/ some impedance equalization
Band Pass Filter - 150Hz and 5KHz w/L-pad
High pass Filter - 7KHz w/ impedance equalization and L-pad

It seems wierd, but this odd combination gives me a +1.29/-5.25 Db. response curve. Actually, it's a nice +/-2.3 Db. from 62 Hz to about 16K; according to the models, at least. The major drawback is that it requires the use of several very large coils. One of which (Lm) needs to be somwhere between 12 and 18 Mh. But the good news is that the DC resistance does not seem to matter much (It's part of the woofer EQ circuit and in series with a resistor anyway) - so I'm looking for cheap 20 ga. coils if I can find any. I want the others to be made of 15 ga or less, if I can afford it.

The Box:
Here's where things get outrageous. It's going to be a huge honkin' tower 51.5"X 19.5" X 22.5" ID. OD will be 53x21x24. Internal volume - accounting for the isobaric chambers and port tubes will be 319 liters. The ports will be tuned at 26Hz. This seems to give me the flattest response with a -3db point at 28.5Hz.

This is one of the reasons I went with the isobaric chamber. Normally someone goes with isobarics to reduce box size or to get slightly extended bass out of a regular box. But I relaized the corollary - even deeper bass response from an infinite baffle box. In order to get the same response out of 2 woofers in a standard configuration, I would need a box twice the size! As it stands, the woofers are the strong point in my models (where the peak is) but I figure I could just reduce my EQ a couple of db at that point. The otherwise flat response far outweighs extra strong bass IMO. I modeled dozens of woofers (from MCM and PartsExpress) and these were actually the flattest among what is within my budget. As it stands, I'm seriously stretching my budget for these monsters. I intend to build 4 of them if the first one turns out to be good enough.

I've played a lot with the port tubes in the simulator and found that 2x20sq. in. area tubes work best (yes, 2 of them) That would equate to about a 5" ID port tube. Since I can't find any, I think I'm going to make them out of aluminum - probably a hexagon or octagon. Anything with a 20 sq in inside area works equallly well as long as one does not get too creative. I start to get some high frequency breakup in the models much beyond that size. I don't know why - perhaps someone can shed some light on the subject. That's another nice thing about a push pull isobaric chamber - there is no noise from the back of the woofer since both backs are inside the chamber. So a front facing port *should* be OK. I essentially don't know any better, so I'm going to try it and see. If anyone knows better, do tell before I screw something up.

I have two major concerns. First is that the bottom of the bandpass filter, at 150 Hz., is really not doing much of anything. It's filtering the worst of the low frequencies out, and the components are keeping the phase from getting too far out of alignment, but at 150Hz, it's just letting the driver's own response curve and Z do most of the work for me. This cant be good for the driver. But if I raise it up, I get a serious drop at the crossover point due to phase shift that get's worse as I increase the filters frequency. Maybe I need a mid-bass unit too, but I can't model anything beyond a 3 way crossover, so I can't predict what would happen if I introduced one or two mid bass drivers. Perhaps if I cascade it off of the bandpass filter? I need some serious advice on this one - Xover pro can't do it for me.

My other concern is phase delay and imaging. Because I had to employ some impedance equalization and a 2nd order filter in the midrange, I had to reverse midrange polarity. As it stands, the mid and woofer are about 18 degrees out of phase at the crossover point (530Hz) (actually it's 378 degrees). I'm not sure if that's much of a problem either. The midrange and tweeter match up quite well, I think.

I've created an acrobat .pdf for anyone interested.
My Project - so far
It's an 89.6K PDF file. I hope the format doesn't bother anyone.
If you have any questions, comments, or criticisms, please feel free to post 'em. Just make it constructive or at least informative, please. I'm just an amateur bordering on noob here and am probably making all the noob mistakes. But I'm too old to be fielding any "Usuk" posts. So I really want to get some second opinions before I commit myself to any rash purchases, if you know what I mean.

Thanks
jdlech

Postcript: I inserted the wrong link in part 1.
Here's the proper link for the high impedance opinion. Just scroll down to the "Technical Sidebar" towards the bottom. I also believe I might get closer to 94-97db sensitivity (maximum) instead of 101. Sorry for any confusion I may have caused.